OpenAI has added a second, more expensive tier to ChatGPT Business. A Premium seat costs $125 a month, or $100 a month with annual billing. The existing Standard seat stays where it was, at $25 a month or $20 annually.
Premium users get five times the usage of a Standard seat and lose the five-hour cooldown that otherwise caps how much work they can push through the tool in one stretch. Businesses can mix both seat types in the same account and hand the pricier one only to the employees who actually need it.
The timing and the price point are not random. Anthropic, OpenAI's closest rival in the business market, already runs the identical structure for its Claude Team plan: $25 for a standard seat, $125 for a premium one. When two competing companies land on the exact same two numbers, that is a sign of what serving a heavy AI user actually costs right now, not a coincidence.
The flat, cheap, unlimited-feeling plans that defined the first few years of ChatGPT and Claude were never built to last. They were priced to get people hooked, not to make money. What changed is how people use these tools.
A quick question typed into a chat box is cheap to answer. An AI agent that pulls sales figures, drafts a customer email campaign, and checks its own output before repeating the process is a completely different cost story. Each extra step runs through the AI model again, and OpenAI pays for every one of those steps in computing power.
As more businesses move from simple chat to these longer, multi-step tasks, the gap between what a light user costs and what a heavy user costs keeps widening. Premium seats are OpenAI's way of finally charging for that gap instead of quietly absorbing it.
There is also a bigger financial story behind this. OpenAI is widely reported to be losing many billions of dollars a year even as its revenue climbs into the tens of billions, and it is doing this while preparing for a possible future stock listing. Investors in that kind of listing want a path toward making money, not just fast growth.
Raising prices on the customers who use the product the most, while leaving casual users untouched, is one of the fastest ways to improve that picture without upsetting the broader user base. Microsoft is running the opposite play with its own Copilot product.
Instead of charging more for the expensive models it buys from OpenAI and Anthropic, Microsoft has started swapping some of those models out for its own in-house ones inside Word, Excel, and Outlook, specifically to cut what it pays per task. Same underlying cost problem, two opposite fixes: OpenAI is passing the cost to the customer, Microsoft is trying to build its way out of the cost.
For any business already paying for ChatGPT Business, the practical move is simple. Look at who on your team is actually running long, multi-step tasks, like building full reports or managing ongoing projects through the tool, versus who is just asking occasional questions.
Only the first group needs the $125 seat, everyone else should stay on the $25 one. It is also worth watching whether Google and Microsoft follow with their own tiered price jumps soon, because flat-rate AI pricing across the industry looks like it is on its way out.